T型三电平储能变流器具有功耗小、通态损耗小、功率密度高等优点, 在低电压大电流场合具有广阔的应用前景。为提升其在动态调节性能与稳态控制精度方面的控制效果, 本文提出的控制方案是利用融合的BP-RBF神经网络建立非线性预测模型, 并在此基础上构建相应的自适应PID控制方法。首先, 本文分析了T型储能变流器的工作原理与数学模型, 并实现电流控制量的解耦与简化。基于此模型, 设计了BP-RBF神经网络, 提出一种支持参数在线调整的自适应PID控制器。通过不同工况下的仿真分析, 证明所提自适应控制策略在响应速度、稳态性能以及整体鲁棒性上均比常规PID控制器具有更好效果, 能够有效抑制输出电流谐波, 提高变流器运行的稳定性与动态性能, 为高性能储能系统的控制策略提供了理论支撑与技术手段。

T-type three-level energy storage converters are widely used in low-voltage, high-current applications due to their low power consumption, low conduction losses, and high power density. To improve their dynamic regulation performance and steady-state control accuracy, this paper proposes an adaptive PID control strategy based on a BP-RBF neural network nonlinear prediction model. First, the operating principle and mathematical model of the T-type energy storage converter are constructed, and the current control variable is decoupled and simplified. Based on this model, a BP-RBF neural network is designed, and an adaptive PID controller with online parameter tuning is proposed. Through simulation analysis under different operating conditions, it is proved that the proposed adaptive control strategy has better response speed, steady-state performance, and overall robustness than conventional PID controllers It effectively suppresses output current harmonics and improves converter stability and dynamic performance, providing theoretical support and technical means for control strategies in high-performance energy storage systems.